1. but loss of consolidation
STM LTM
Retrieval
STM LTM
STM LTM
Storage; Consolidation
Case of H. M.
Hippocampus
Not loss of LTM,
Not loss of STM,
but loss of consolidation

2. Tolman: Cognitive MapsSt G

3. How? Intramaze (e.g., smell food, scratch on floor)
Olton: Radial Arm Maze
How?
Intramaze (e.g., smell food, scratch on floor)
Extramaze (stimuli around the maze)

4. Choose 4
Then rotate maze * * * *

5. Which will they visit?
Go to “space”, not arm
Use extramaze cues * * * *

6. Morris: Water Maze
Hence:
“Spatial”, and not any single cue.

7. Morris: Water Maze
Further tests:
Probe trial

8. Role of Hippocampus in Spatial Learning
Hippocampal Damage:
Still get a little faster at finding platform
Probe Trial:
Shows no “spatial strategy

9. Role of Hippocampus in Spatial Learning
If visible platform, do as well as normal
Suggests role of hippocampus specifically in the “spatial” aspects of the water maze task

10. Configural Learning Hypothesis
Spatial Learning is type of configural learning:
response is based on > 1 cue
Hippocampus and Configural learning:
If Hippocampal damage, can do:
discrimination,
but not configual
A—US
B—US
AB—
A—US B—

11. Integrating across experiences
Rule learning (learning set)
Strategies
Win-stay
Win-shift
Strategy is then applied to new situations:
More rapid learning of new (same strategy) task.

12. Hypothesis Testing
New task: Try different hypotheses (strategies)
If wrong, throw it out
If correct, stick with it
Leads to “step function” learning curves
looks like “insight”
Percent
Correct
Trials
100

13. Concept Formation
Concept: Category- a set of objects/events having common features/relationships
Abstraction: How a concept is formed Abstract (summarize) common features.
Form prototype (composite average) based on experiences with specific exemplars (examples)

14. “Neural networks” and prototype formation
Common assumptions:
1. Universal connectivity
(note: preparedness: perhaps not “universal”)
2. Contiguity
3. Spread of activation

15. “Neural networks” and prototype formation
Common assumptions:
1. Universal connectivity
2. Contiguity
3. Spread of activation
4 legs
ears
tail
fur

16. Learning rule: Delta Rule
4 legs
ears
tail
External input
fur
Internal input

17. Learning rule: Delta Rule
Sensory neuron
Sensory neuron
External input T E
Internal input
∆ synaptic efficacy
∆ Int
= c (Ext - Int)
REM: ∆V = c (Vmax - V)

18. Get Expect Similar Ideas External Internal Vmax V Sensory detection
and activation Retrieval
of a node
Difference = Surprise